Physical Reasoning Using Dynamics-Aware Models

TL;DR

This paper introduces a dynamics-aware modeling approach that enhances physical reasoning by incorporating self-supervised signals about object trajectories, leading to improved performance on the PHYRE benchmark.

Contribution

It proposes augmenting reward signals with self-supervised object dynamics to improve physical reasoning models, achieving state-of-the-art results.

Findings

Significant performance improvements on PHYRE benchmark

Self-supervised signals enhance physical reasoning accuracy

Trajectory similarity measures improve model training

Abstract

A common approach to solving physical reasoning tasks is to train a value learner on example tasks. A limitation of such an approach is that it requires learning about object dynamics solely from reward values assigned to the final state of a rollout of the environment. This study aims to address this limitation by augmenting the reward value with self-supervised signals about object dynamics. Specifically, we train the model to characterize the similarity of two environment rollouts, jointly with predicting the outcome of the reasoning task. This similarity can be defined as a distance measure between the trajectory of objects in the two rollouts, or learned directly from pixels using a contrastive formulation. Empirically, we find that this approach leads to substantial performance improvements on the PHYRE benchmark for physical reasoning (Bakhtin et al., 2019), establishing a new state-of-the-art.